GS-SDF: LiDAR-Augmented Gaussian Splatting and Neural SDF for Geometrically Consistent Rendering and Reconstruction

1. Introduction

A unified LiDAR-visual system achieving geometrically consistent photorealistic rendering and high-granularity surface reconstruction. We propose a unified LiDAR-visual system that synergizes Gaussian splatting with a neural signed distance field. The accurate LiDAR point clouds enable a trained neural signed distance field to offer a manifold geometry field. This motivates us to offer an SDF-based Gaussian initialization for physically grounded primitive placement and a comprehensive geometric regularization for geometrically consistent rendering and reconstruction.

Our paper is currently undergoing peer review. The code will be released once the paper is accepted.

Project page | Paper | Video

2. Related paper

GS-SDF: LiDAR-Augmented Gaussian Splatting and Neural SDF for Geometrically Consistent Rendering and Reconstruction

FAST-LIVO2: Fast, Direct LiDAR-Inertial-Visual Odometry

If you use GS-SDF for your academic research, please cite the following paper.

@article{liu2025gssdflidaraugmentedgaussiansplatting,
      title={GS-SDF: LiDAR-Augmented Gaussian Splatting and Neural SDF for Geometrically Consistent Rendering and Reconstruction}, 
      author={Jianheng Liu and Yunfei Wan and Bowen Wang and Chunran Zheng and Jiarong Lin and Fu Zhang},
      journal={arXiv preprint arXiv:2108.10470},
      year={2025},
}